Using deep learning in the detection of breast cancer in histopathological imaging
Machine learning has massive implications for the healthcare sector. Advances in computer vision and deep learning can allow us to automate the time-consuming and costly nature of manually annotating metastases in histological imaging. Patch extraction is especially dominant in the field when making attempts at managing very large Whole Slide Images and is generally considered a necessity due to the computational load of processing images hundreds of thousands of pixels in width and height. This project looks to create a patch classifier for Whole Slide Images in the detection of breast cancer in histological images of lymph node sections. Furthermore, an analysis of the effects of hyper-parameters and pre-processing methods on the output accuracy of the deep learning model was undertaken using a variety of investigations and experiments.
Keywords
Deep Learning, Neural Networks, Machine Learning, Computer Vision, Imaging, Medical, Cancer, Classifier, Pathology, Histopathology
Staff
[David Harris-Birtill]{dcchb}